Projects / Programmes
Measuring psychophysiological parameters as input data for computerized adaptive testing
| Code |
Science |
Field |
Subfield |
| 2.15.04 |
Engineering sciences and technologies |
Metrology |
Metrologies in areas |
| Code |
Science |
Field |
| P180 |
Natural sciences and mathematics |
Metrology, physical instrumentation |
| Code |
Science |
Field |
| 2.02 |
Engineering and Technology |
Electrical engineering, Electronic engineering, Information engineering |
Measurements, psychophysiology, computerised adaptive testing, accuracy, uncertainty, psychophysiological feedback
Researchers (12)
Organisations (4)
Abstract
In psychometrics, which deals with measurement of human psychological characteristics, computerized adaptive testing is frequently used. In this type of testing, the computer selects items according to participant’s responses to previous items, so that the measurement error and uncertainty of test score are minimized. In this procedure, the emotional-motivational aspects of testing are ignored, but they might affect the test performance of individuals. The ultimate goal of our project is to measure physiological responses during testing and, based on these measurements, to determine how the individual experiences the testing situation and to choose an item of appropriate difficulty accordingly. The goal is also to steer the process of the testing in such a way that maximum satisfaction and comfort of the individual during the testing will be achieved. We will investigate whether adaptive testing that adjusts according to the estimate of emotional-motivational state of the participant based on psychophysiological measures, gives comparable estimates of participant’s ability and standard error of estimate as a conventional computerized adaptive testing, and leads to a comparable overall satisfaction of the participant with the testing.
The project will first identify the ability to be measured, and a large pool of items will be developed. Their psychometric characteristics will be examined on a large pilot sample according to item-response theory procedures.
We will develop a methodology for determining the basic metrological parameters of psychophysiological measurements. Using classical and non-intrusive psychophysiological instrumentation we will yield information about skin conductance, heart rate, blood pressure and respiratory rate during solving specific items, and search for those psychophysiological features and their combinations that will be able to differentiate between the tasks in which the participant is experiencing pleasure (tasks of appropriate difficulty), tasks in which the participant is bored (tasks that are too easy), and tasks in which she/he is experiencing mental stress and discomfort (tasks that are too difficult).
Then three experiments will be carried out in which we will compare the ability estimates based on the correctness of task solutions and ability estimates based on psychophysiological measures and the estimate of the emotional-motivational state of the participant during testing. Both types of ability estimates will be compared (i) for a non-adaptive test procedure, (ii) for the conventional computerized adaptive testing, where the next item in a sequence is selected based on participant’s responses, and (iii) the adaptive procedure guided by a recommendation system that selects the next item on the basis of the estimate of participant’s emotional-motivational state derived from the psychophysiological measures. We will also examine whether the individual satisfaction with the test procedure that adapts to the emotional-motivational state is larger than with the testing procedure which adapts according to the participant’s responses.
The original contributions of the project will be the following: (i) we will propose a concept of metrological evaluation of psychophysiological measurement devices and (ii) we will include the indicators of the emotional-motivational state of a participant into adaptive psychological testing and evaluate such a new form of adaptive testing.
Significance for science
Currently, no systems of computer adaptive testing (CAT) incorporating physiological measurements exist. The reason, in addition to natural physiological variability of physiological parameters and the difficulty of determining the criteria function for decision, is the intrusiveness of classical measuring of psychophysiological parameters. The project is the first of its kind, exploring the possibility of a physiological CAT. The development of non-intrusive techniques of psychophysiological measurements would mean the possibility of using physiology in different areas, for example in medical diagnostics (therapy, clinical psychology), environmental and sports medicine (well being of an athlete), gerontology (security systems, smart home for the elderly), educational sciences (effectiveness of learning, psychological state of students and teachers, comparison of different learning and teaching methods, adaptive textbooks) and in decision-making of recommender systems (prediction of users decision based on personal profiles), etc. In metrological terms the importance of the results of the research project is in laying the foundations of new forms of measurement instrumentation, which are highly multidisciplinary (physiological measurement, psychometrics). Classical metrology principles (error, uncertainty) were introduced in the field of psychophysiological instrumentation and the results published. Experience gained in the project will serve as the foundation of future studies of making new psychodiagnostic instruments. Further, the project has led to new insights into advanced form of computer adaptive testing. The concept of the system, which in addition to the classic principles of testing (taking into account the student's answers) also uses hers/his physiology and uses them in an adaptive feedback loop, is in the current stage of research still impossible in real time. But we gained bits on the characteristics of the feedback system, which would be functional. A special measurement protocol would be needed that would enable gathering of a large enough quantity of data within each individual, so that physiological responses of an individual during solving tasks of different difficulty would be more reliably predicted, using with the procedures related to per-person feature normalisation.Inclusion of psyciological measures as indicators of emotional-motivational state of the user could lead to high user satisfaction and, consequently, to a higher quality of psychodiagnostic services. In this way, the general detection of user situation by means of physiological sensors and predictive models of machine learning would mean that there would be no need to break up the test for the filling-in the questionnaires or providing subjective ratings of the current state or experience. Users would be able to remain attentive to the content of the test, which would in turn enable to collect a larger amount of data in the same time or shorten the testing time.
Significance for the country
Based on the experiences gained in the project we can suggest ways of processing physiological signals and features in further research of physiological responses to cognitive load. The study showed that on the group level we can validly use physiological physiological responses for monitoring cognitive load in solving cognitive tasks. Physiological measures will therefore be useful in future studies of various cognitive processes for monitoring the changes in cognitive, emotional and motivational state of the participant during long measurements. Physiological measures seem to be useful for evaluating the effects of cognitive trainings. As a non-invasive measures of cognitive load they are also useful in clinical practice or with populations that have difficulties switching between mental activities during solving test tasks and during subjective assessment of the task difficulty. The use of developed instruments for non-intrusive measurement of physiological responses can be broad. For example, they can be used in research and clinical practice (for monitoring patients during treatment, athletes during training, learners during learning, etc.).
Most important scientific results
Annual report
2013,
2014,
2015,
final report
Most important socioeconomically and culturally relevant results
Annual report
2013,
2014,
2015,
final report
